about modeling

© All Rights Reserved

17 views

about modeling

© All Rights Reserved

- Facimage Mrgc Paper
- Regression Checklist
- 0120-IdentifyingDuplicateRecordsinIDQ9
- Fin Sol
- biochemj00812-0107
- 5832_chap1
- About the MS Regress Package
- Dividend Changes and Future Profitability
- Financial Leverage and Performance of Non-Financial Companies in Nairobi Securities Exchange in Kenya
- Statistics study notes
- TR-2003011
- evaluacion affordances
- tmpCC92.tmp
- Xl Miner
- Upload
- Chapter17_3
- Cluster Methods in SAS
- paper_tit
- bettcher term paper apsy 607
- ewan

You are on page 1of 12

Carbonate Reservoir

Farhad Khoshbakht

Department of Reservoir Study and Field Development, Research Institute of Petroleum Industry (RIPI), Tehran,

Iran

ABSTRACT

Permeability can be directly measured using cores taken from the reservoir in the laboratory. Due

to high cost associated with coring, cores are available in a limited number of wells in a field. Many

empirical models, statistical methods, and intelligent techniques were suggested to predict

permeability in un-cored wells from easy-to-obtain and frequent data such as wireline logs. The

main objective of this study is to assess different approaches to the prediction of the estimation of

permeability in a heterogeneous carbonate reservoir, i.e. Fahliyan formation in the southwest of

Iran. The considered methods may be categorized in four groups, namely a) empirical models

(Timur and Dual-Water), b) regression analysis (simple and multiple), c) clustering methods like MRGC

(multi-resolution graph-based clustering), SOM (self organizing map), DC (dynamic clustering) and AHC

(ascending hierarchical clustering), and d) artificial intelligence techniques such as ANN (artificial

neural network), fuzzy logic, and neuro-fuzzy.

This study shows that clustering techniques predict permeability in a heterogeneous carbonate

better than other examined approaches. Among four assessed clustering methods, SOM performed

better and correctly predicted local variations. Artificial intelligence techniques are average in

modeling permeability. However, empirical equations and regression methods are not capable of

predicting permeability in the studied reservoir. The constructed and validated SOM model with

69 clusters was selected to predict permeability in the blind test well of the studied field. In this

well, the predicted permeability was in good agreement with MDT and core derived permeability.

costs and technical limitations. Permeability as

one of the most important petrophysical

properties of a reservoir is the main input to

static and dynamic models. In order to construct

a realistic model of a reservoir, accurate

permeability data with an appropriate distribution

ought to be available. Thus it is essential to

predict permeability from easy-to-obtain and

INTRODUCTION

In addition to being porous, a reservoir rock

must be able to allow fluids to flow. The ability

of rock to conduct fluids is defined as

permeability. Core is the most reliable source of

matrix permeability. Usually in a field, core is

available in a limited number of wells because of

*Corresponding author

Article history

Farhad Khoshbakht

Email: khoshbakhtf@ripi.ir

2014 Research Institute of Petroleum Industry (RIPI)

79

http://jpst.ripi.ir

Journal of Petroleum

Science and Technology

permeability is not straightforward. The difficulty

of predicting permeability is due to the fact that

carbonate rocks are very heterogeneous and

factors such as diagenesis (e.g. cementation),

texture, grain size variation and fabric, cementation,

and clay content control permeability [10]. Unlike

core, wireline logs are available in almost all wells in

a field. By establishing a robust model which relates

permeability to wireline log responses permeability

in un-cored wells can be predicted.

More than 40 percent of the world oil reservoirs

are placed in carbonate rocks. Because of the

economic importance of carbonates, their

accurate characterization is vital for exploration

and production. The ultimate goal of reservoir

characterization is to improve production efficiency

and oil recovery by understanding and modeling

the reservoir. To correctly model the flow

behavior in a carbonate reservoir, it is essential

to understand the permeability profile [20].

Carbonate rocks are unstable and diagenetic

processes chemically and physically alter a

somehow homogeneous formation into a

completely heterogeneous one. The permeability of

a carbonate reservoir often is severely affected by

tectonic and diagenetic phenomena like fracturing,

dissolution, and leaching, which makes the

prediction of permeability very difficult.

In the last decade, many researchers have tried

to estimate permeability from wireline logs.

Because of availability and frequency of wireline

logs, they have become a popular source for

estimating permeability. Archie (1942) introduced

relationships which estimated permeability via

core analysis data such as porosity and formation

resistivity factor [3]. Porosity, water saturation,

capillary pressure, formation resistivity factor,

and NMR T1 & T2 parameters, derived from

wireline logs, are used to estimate permeability.

Leverett (1941), Tixier (1949), Wyllie and Rose

(1950), Timur (1968), and Coates and Dumanoir

F. Khoshbakht

measurements to determine permeability [5,

21-23].

In recent years, artificial intelligence techniques

have widely been used to estimate permeability

from wireline logs. Balan et al. (1995),

Mohaghegh et al. (1997), Lin et al. (1994), Zhang

et al. (1996), and Huang et al. (1996) predicted

permeability by means of neural networks.

Cuddy (1998) and Finol et al. (2001) used the

fuzzy approach to estimate permeability [4,79,13,16,25].

In this study, different approaches to permeability

prediction are examined in order to establish a

robust model to predict permeability in un-cored

wells. The method with the best performance,

i.e. SOM, was selected to predict permeability in

un-cored wells. The predicted permeability in

the blind test wells was compared with MDT(Modular Formation Dynamics Tester) derived

permeability and core permeability to ensure

that the model worked properly.

MATERIALS AND METHODS

The studied field is located in Zagros fold-thrust

belt, the southwest of Iran (Figure 1). Zagros is a

part of Tethys Ocean and is one of the most

important petroleum basins in the world [1].

This basin is located in the south west of Iran

and north of Iraq. The geological history of this

basin includes long time subsidence and deposition

interrupted by short time uplift. Folding process of

this basin occurred in Miocene and Pliocene and

continued until now, which has formed long

anticlines [17]. These anticlines constitute most

of oil traps in this basin.

Fahliyan (the uppermost Jurassic-Cretaceous) is

one of the major reservoir formations of Iranian

oil fields. Fahliyan is a clean limestone comprised

of massive oolitic or pelletal limestone [2]. In the

studied field, Fahliyan formation is divided to

eight reservoir zones.

2014 Research Institute of Petroleum Industry (RIPI)

80 |

http://jpst.ripi.ir

Journal of Petroleum

Science and Technology

and lithology.

selected wireline logs (From left to right: NHPI, DT,

LLD, PEF, and GR)

Permeability Modeling

wells obtained from Fahliyan formation were

used. The datasets of three wells (A, B, and C)

were used for building models and the forth well

(D) left out for the test. A suit of logs including

neutron porosity (NPHI), sonic log (DT),

resistivity log (LLD), photoelectric (PEF), and

natural gamma ray (GR) was selected as the

input to all the models (Figures 2 and 3). These

logs record the properties of the reservoir,

which control permeability such as shale

wells, models should be constructed. The inputs

to these models are well logs and the output is

permeability. In wells with core data (A, B, and

C), the models were constructed and tuned, and

then they were used to predict permeability in

un-cored wells. In the following, all the models

are briefly described and the predicted

permeability by means of each one is presented.

Figure 3: Statistics and histogram of permeability as the output of the models and selected wireline logs

(NHPI, DT, LLD, PEF, and GR) as the inputs to the prediction models

Journal of Petroleum Science and Technology 2015, 5(1), 79-90

2014 Research Institute of Petroleum Industry (RIPI)

http://jpst.ripi.ir

| 81

Journal of Petroleum

Science and Technology

F. Khoshbakht

Regression Analysis

Empirical Models

Empirical models are based on experiential

observations and are introduced for particular

reservoirs. For applying them to other reservoirs,

with different properties, it is necessary to

calibrate their parameters. Generally empirical

equations relate permeability to porosity and

water saturation. First, we calibrated the models

by using the core data of train wells. In train

wells, the inputs (porosity and irreducible water

saturation) and the output of the model

(permeability) were available and the only unknown

of the equations (A and B) can be calculated. Then,

in blind-test well the permeability was predicted

using the calibrated models. In this study, we

used two common empirical equations for

predicting permeability: Timur and Dual-Water.

Timur (1968) found a relationship for estimating

the permeability of sandstones from the in-situ

measurements of porosity and residual saturation

[21]. His model is applicable where the condition

of residual saturation exists. The main source of

uncertainty in Timur model is the high amount

of error in determining the residual saturation.

In our case, the residual water saturation of the

studied reservoir was obtained from capillary

pressure tests. The average value of residual

saturation was 0.18. The Dual water model was

developed by Coates and Dumanoir in 1974 [5].

They suggest this model for shaly formations.

Total porosity, effective porosity, and irreducible

water saturation are the inputs to this model.

The equations and calibrated parameters are

presented in Table 1.

Perm = 10(

identifies the relationship between two or more

quantitative variables: a dependent variable whose

value is to be predicted, and independent

variable(s). Regression analysis is used to

understand the statistical dependence of one

variable on other variables.

We utilized simple and multiple regressions in

order to estimate permeability from wireline

logs. In the simple regression, the effective

porosity (PHIE) was the input (independent

variable). In the multiple regression, DT, LLD,

NPHI, PEF, and GR were selected as the inputs.

In the multiple regression, the data is

constituted using the below equation:

( perm ) = a0 + a1 ( DT ) + a2 ( LLD )

a3 ( NPHI ) + a4 ( PEF ) + a5 (GR )

Log

coefficients (a0, a1, a2, a3, a4, and a5) are

unknown. A system of matrix (Y=aX) representing

the above linear equations is constructed and

solved for unknown coefficients by minimizing

the sum of the squares of the deviations of the

data from the model (least-squares fit). The

equation below represents the relationship

between permeability and PHIE, which is

obtained from the simple regression.

= 10(

dependent variable and well logs as the independent

variables is given by the below equation:

1.748+ 0.00884115( DT ) 0.132608 Log ( LLD ) +1.64347( NPHI ) +385849( PEF ) + 0.00575193( GR ) )

Table 1: Used empirical equations and their parameters calibrated for the studied wells

Name

Equation

Timur

Dual-Water

=& .

'.

!"##

)!"## %

*

)!"##

Constant (A)

Exponent (B)

R2 of prediction

6.758595

1.230943

0.6028

8.001080

1.176190

0.5787

2014 Research Institute of Petroleum Industry (RIPI)

82 |

http://jpst.ripi.ir

Clustering Methods

The aim of cluster analysis is to classify a dataset

into groups that are internally homogeneous

and externally isolated on the basis of a

measure of similarity or dissimilarity among

groups [12]. Clustering methods are widely used

for electro-facies analysis and the prediction of

petrophysical properties. By integrating clustering

methods with intelligence techniques, some new

methods such as SOM and MRGC were born

[11,24]. In this study, we utilized two classic

clustering methods, namely DC and AHC, and

two new clustering methods, i.e. SOM and

MRGC, to predict permeability from wireline

logs.

Predicting permeability by means of clustering

methods is carried out in three steps:

1) Partitioning log data of the train wells into

somehow similar groups and determining the

properties of each groups;

2) Identifying clusters of test data using cluster

properties, determined in the previous step;

3) Predicting permeability in each cluster of the

blind-test well by means of KNN method.

Each clustering method tries to partition data

into stable and homogeneous groups on the

basis of its particular algorithm. Selecting the

optimal number of clusters is important, because

variation in the number of clusters strongly affects

the results. There are no universal criteria for

choosing the optimal number of clusters in the

examined methods. Therefore, we tried

different values for the number of clusters and

chose the optimal values via trial and error.

NPHI, DT, LLD, PEF, and GR were the inputs to all

the clustering methods.

MRGC is based on non-parametric K-nearest

neighbor and graph data representation. MRGC

is a tool which analyzes the structure of the

complex data and partitions natural data groups

into different shapes, sizes, and densities [24].

Journal of Petroleum

Science and Technology

number of clusters. We assigned 4 and 35

respectively as the minimum and maximum of

the desired clusters for MRGC. After executing

MRGC, the log data of train wells were

partitioned into six orders of clusters (5, 7, 10,

14, 16, and 18 clusters). The highest R2 (0.5978)

was obtained when MRGC model with 16

clusters was used for permeability prediction

(Figure 4-a).

The SOM is a computational method for the

visualization and analysis of high-dimensional data,

especially experimentally acquired information

[11]. We tested SOM networks with different

dimensions to partition the datasets of train

wells. Then, the permeability of the blind-test

well was predicted using each model (Figure 4b). As shown in Figure 4-b, the optimal dimension

of SOM network is 69, which gives the highest R2

(0.6936) (Table 2 and Figure 5-f).

Dynamic Clustering or k-means clustering is a

non-hierarchical method to classify a dataset on

the basis of a pre-defined number of clusters.

The algorithm assigns each object into a cluster

by iteration. It minimizes the sum of distance

from each object to its cluster centroid, over all

clusters. This algorithm moves objects between

clusters until the sum cannot be decreased

further [15]. This clustering method is suitable

for datasets with a large amount of data. We

examined different values for the number of

clusters and predicted permeability in the blindtest well. Permeability prediction on the basis of

12 clusters (Figure 4-c) has the highest R2 (0.6631)

among DC models with different numbers of

clusters (Figure 5-g).

Hierarchical clustering method partitions data

over a variety of scales by creating a cluster tree

or dendrogram [15]. Ascending hierarchical cluster

(AHC) analysis is a statistical method for finding

relatively homogeneous clusters of cases based

on measured characteristics [14]. AHC produces

different levels of clusters and user can choose

2014 Research Institute of Petroleum Industry (RIPI)

http://jpst.ripi.ir

| 83

Journal of Petroleum

Science and Technology

F. Khoshbakht

Generally, clustering of a dataset can be

executed by means of AHC in three steps: 1)

measuring the distance between every pairs of

objects in the dataset, 2) linking pairs of objects

which are very close together, and 3) cutting the

hierarchical tree into clusters [15]. Hierarchical

trees were cut at different levels and the results

showed that AHC model with 8 clusters has the

highest R2 (Figure 4-d) among levels of clusters.

AHC models with less or more than 6 clusters

decrease the R2 value (Figure 5-h). Increasing

the number of clusters does not make sharp

changes in the results, because AHC combines

low order clusters.

Artificial Intelligence Techniques

ANN, fuzzy logic, and neuro-fuzzy are three

artificial intelligence techniques which were

used in this study to predict permeability from

wireline logs.

Neural network is a modeling technique which

models systems in a brain-like way. The main

feature of neural network is that it can learn the

internal characteristics of a system by analyzing

datasets.

MRGC

SOM

0.70

0.70

0.65

0.65

R2

R2

0.60

0.60

0.55

0.55

0.50

0.50

7

10

14

16

18

No. of Clusters

89

78

68

59

56

48

45

38

35

29

26

23

88

55

22

Dimension of Network

AHC

Dyn. Clust.

0.70

0.70

0.65

0.65

R2

R2

0.60

0.60

0.55

0.55

0.50

0.50

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

No. of Clusters

10

11

12

No. of Clusters

2

Figure 4: Variation of R by changing the number of clusters for a) MRGC, b) SOM, c) DC, and d) AHC

2014 Research Institute of Petroleum Industry (RIPI)

84 |

http://jpst.ripi.ir

back propagation neural network with two

hidden layers have been designed. A trained

network was used to predict the permeability of

the test well. The R2 of the predicted permeability

by means of neural network versus core

permeability is 0.6279 (Table 2 and Figure 5-i).

In this paper, a fuzzy approach proposed by

Cuddy (2000) was utilized for permeability

prediction [7]. This approach works by assigning

a probability to the quality of the prediction

from each log, then combines the probabilities,

and predicts the most likely permeability [7]. We

examined different numbers of classes and

predicted the permeability for each case. Fuzzy

model with 10 classes predicted permeability

with the highest R2 (0.6275) among models with

different numbers of classes (Figure 5-j).

We used a locally linear neuro-fuzzy (LLNF)

model which utilized local linear model tree

learning algorithm (LOLIMOT) to predict the

permeability. The fundamental approach to a

locally linear neuro-fuzzy model is dividing the

input space into small linear subspaces with

fuzzy validity functions. Any produced linear

part with its validity function is described as a

fuzzy neuron. Thus the total model is a neurofuzzy network with one hidden layer, and a

linear neuron in the output layer, which simply

calculates the weighted sum of the outputs of

locally linear models [18]. By plotting the mean

square error for the train and test data, we

found that the optimal number of hidden layer

neurons is 29, and constructed our neuro-fuzzy

model with 29 neurons. The R2 of permeability

prediction by means of neuro-fuzzy model was

0.5900 (Figure 5-k).

RESULTS AND DISCUSSION

The prediction of permeability in un-cored wells

is one of the difficult tasks in reservoir

characterization, especially in heterogeneous

Journal of Petroleum

Science and Technology

frequent and easy-to-obtain data, which are

usually available in almost all the wells. We

launched this study to examine four common

approaches to predicting permeability, namely

empirical methods, statistical analysis, clustering

methods, and artificial intelligence techniques.

The method with the best performance was

selected to predict permeability in un-cored

wells. The predicted permeability in one of the

un-cored wells of the studied field was

compared with MDT- (Modular Formation

Dynamics Tester) derived permeability to ensure

that the model worked properly.

The first examined approach was empirical

equation: Timur and Dual-Water models. The R2

of the measured permeability versus predicted

permeability using Timur and Dual-Water models

are 0.6028 and 0.5787 respectively (Table 2).

Even though the predicted permeability by

means of Timur model is closer to the measured

permeability than the one measured by DualWater model, both models performed weakly

(Figures 5-a and 5-b). In spite of our efforts to

properly calibrate the parameters of these

models, the permeability prediction was not

satisfactory. Because of the heterogeneity of the

reservoir, empirical equations failed to predict

permeability properly. Usually the use of

empirical equations is restricted to the region(s)

where the models are developed for, and in

different reservoirs, imperial models should be

employed with caution.

Regression analysis is simpler and easier to

practice than other methods. The R2 values of

permeability prediction by means of simple and

multiple regressions are 0.4920 and 0.5279

respectively (Table 2). The distribution shape of

the predicted data via simple and multiple

regressions are similar and both overestimate

low permeabilities and underestimates high

permeabilities.

2014 Research Institute of Petroleum Industry (RIPI)

http://jpst.ripi.ir

| 85

Journal of Petroleum

Science and Technology

F. Khoshbakht

Table 2: The R2 of the predicted versus measured permeability for all the used methods

Type

Model

R2

Empirical

Regression

Clustering

DualWater

Simple

Multiple

MRGC

16

0.6028 0.5787

0.4920

0.5279

Timur

(Figure 2) show that there is not a simple and

linear relationship between them. Consequently, we cannot expect methods like

empirical equations and regression analysis to

explore complex relationships. Empirical

methods simplify problems by reducing the

number of variables, while different factors

influence the permeability of reservoirs. In

addition, regression methods tend to average

data and ignore local variations, which are

important in thin layered reservoirs. Figures (5c) and (5-d) present the plots of the predicted

permeability versus the measured permeability

by regression analysis.

The aim of cluster analysis is to classify a dataset

into groups that are internally homogeneous

and externally isolated on the basis of a

measure of similarity or dissimilarity between

groups [12]. For proper permeability prediction

in complex reservoirs, different criteria have

been proposed for data classification such as

reservoir layers, litho-facies, rock types, and

hydraulic flow units. But in order to predict

permeability from logs, methods should be

employed which can handle the dimensionality

and complexity of log data. Because of the

variety of wireline logs and the heterogeneity of

carbonates, it is usually difficult to easily

recognize natural groups of data.

SOM

69

DC

12

Artificial Intelligence

AHC

6

ANN

Fuzzy 10 Neuro-Fuzzy

0.5900

acceptable. Selecting the optimal number of

clusters is one of the important tasks in using

clustering techniques, because variation in the

number of clusters strongly affects the result of

permeability prediction. In addition, there are

no general criteria for choosing the optimal

number of clusters in the examined methods.

We examined different values for the number of

clusters and predicted permeability in each case.

Each method has its own value for the optimal

number of clusters, e.g. the optimal number of

clusters for MRGC, SOM, DC, and AHC are 16,

54, 12, and 6 respectively. Predicting permeability

on the basis of SOM has the highest R2 among the

examined clustering methods, even though SOM

is not much more robust than the other

clustering methods; but, the results of the

prediction are slightly comparable. However,

SOM predicts low and high permeabilities better

than the other methods. The performance of

ANN and fuzzy logic for predicting permeabilities

between 1 to 10 mD is acceptable, but both

methods fail to predict permeabilities higher

than 10 mD correctly. Contrary to our expectation

neuro-fuzzy performance was weaker than the

performance of ANN and fuzzy logic. Neurofuzzy only predicts very high values (~100 mD)

properly, but for the range 1 to 10 mD, at which

most methods do well, neuro-fuzzy was

disappointing.

better than the other examined methods. Even

2014 Research Institute of Petroleum Industry (RIPI)

86 |

http://jpst.ripi.ir

Journal of Petroleum

Science and Technology

Figure 5: Cross-plots of the measured permeability versus the predicted permeability by means of different

methods, the methods and the R2 of the prediction are presented in the up-left box of each cross plot.

their ability to predict local variations. Most

examined methods, especially regression and

empirical methods, ignore local variations. These

methods predict medium permeabilities well, but

the prediction of extremes is associated with high

errors. Statistically the number of samples in the

extremes is low and most techniques cannot

learn the pattern of these parts of data in

training. Consequently, they tend to average the

data and ignore local variations.

The best performed method (SOM with 69

of un-cored wells of the studied field. In this

well, the permeability derived from MDT test

was available. As it is clear in the right column of

Figure 6, the predicted permeability matches

well with MDT permeability. This interval

comprises limestone with a limited amount of

shale and its porosity is almost constant. Utilizing

empirical methods which relate permeability to

porosity leads to somehow constant predicted

permeability, assuming that irreducible water

saturation is also constant in this interval.

2014 Research Institute of Petroleum Industry (RIPI)

http://jpst.ripi.ir

| 87

Journal of Petroleum

Science and Technology

F. Khoshbakht

Figure 6: Wireline logs and predicted permeability in the blind test well; From left: Track 1) CGR and GR,

Track 2) depth scale, Track 3) lithology, Track 4) resistivity logs, Track 5) NPHI, DT & RHOB, Track 6) predicted

permeability by means of SOM 69 (solid line) and MDT permeability (dots), Track 7) predicted permeability

by means of SOM 69 (solid line) and measured permeability (square).

2014 Research Institute of Petroleum Industry (RIPI)

88 |

http://jpst.ripi.ir

Journal of Petroleum

Science and Technology

CONCLUSIONS

Numerous methods have been proposed by

researchers to predict permeability in un-cored

wells. Common approaches to predicting

permeability are empirical models, regression

analysis, clustering methods, and artificial

intelligence techniques. We examined these

methods to predict permeability in a carbonate

reservoir. The study showed that among the

examined methods, regression analysis was the

worst and clustering techniques were the best in

predicting permeability. Accordingly, among the

four assessed clustering methods, SOM performed

better and could properly predict local variations.

Since the clustering methods classify the datasets

into homogeneous subclasses and then predict the

permeability in each sub-class, it is expected that

the clustering methods perform better compared

with the other methods which treat the dataset as

a whole; in fact, in each subclass the relationship

between permeability and the log responses is

simpler.

Artificial intelligence techniques were average in

modeling permeability. Empirical equations and

regression techniques were not capable of

predicting permeability in the studied reservoir.

The successful prediction of permeability in an

un-cored well, confirmed by MDT, justified the

capability of SOM as a clustering technique.

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

REFERENCES

[1] Alavi M., Tectonics of the Zagros

Orogenic Belt of Iran: New Data and

Interpretations, Tectonophysics, 1994,

229, 211-238.

[2] Alavi M., Regional Stratigraphy of the

Zagross Fold-thrust Belt of Iran and its

Proforeland Evolution, American Journal

of Science, 2004, 304, 1-20.

[3] Archie G. E., The Electrical Resistivity Log

as an Aid in Determining Some Reservoir

Characteristics, Petroleum Transactions,

1941.

[4] Balan B., Mohagheg S., and Ameri S., State

of the Art in Permeability Determination

[13]

[14]

[15]

[16]

2014 Research Institute of Petroleum Industry (RIPI)

Model Development, paper SPE 30978

presented at SPE Eastern Regional Conference

& Exhibition held in Morgantown, West

Virginia, U.S.A, 1995.

Coates G. R. and Dumanoir J. L., A New

Approach to Improved Log Derived

Permeability, The Log Analyst, 1974.

Cuddy S. J., The Application of the

Mathematics of Fuzzy Logic to the

Geosciences, Paper SPE 49470, 1998.

Cuddy S. J., Litho-Facies and Permeability

Prediction from Electrical Logs Using Fuzzy

Logic, paper SPE 65411, 2000.

Finol J., Guo Y. K., and Jing X. D., A Rule

Based Fuzzy Model for the Prediction of

Petrophysical Rock Parameters, Journal

of Petroleum Science and Engineering,

2001, 29, 97-113.

Huang Z., Shimeld J., Williamson M., and

Katsube J., Permeability Prediction With

Artificial Neural Network Modeling in the

Venture Gas Field, Society of Exploration

Geophysics, 1996, 61(2), 422-436.

Jennings J. W. and Lucia F. J., Predicting

Permeability from Well Logs in Carbonates

with a Link to Geology for Interwell

Permeability Mapping, Paper SPE 84942PA, 2003.

Kohonen T., Self-Organizing Maps, 3th

ed, Springer, 2001, 501.

Lee S. H. and Datta-Gupta A., Electrofacies Characterization and Permeability

Predictions in Carbonate Reservoirs: Role of

Multivariate Analysis and Nonparametric

Regression, Paper SPE 56658, Houston,

Texas, 1999.

Lin J. L. and Salisch H. A., Determination

from Well Logs of Porosity and Permeability

in a Heterogeneous Formation, paper SPE

28792, presented at the SPE Asia Pacific Oil

& Gas conference, Melbourne, Australia,

1994.

Lukasova A., Hierarchical Agglomerative

Clustering Procedure, Pattern Recognition,

1979, 1, 365-381.

Matlab, Statistics Toolbox users Guide,

Mathworks, Inc, 2007, 1561.

Mohaghegh S., Balan B. and Ameri S.,

Permeability Determination from Well

Log Data, SPE International Journal,

http://jpst.ripi.ir

| 89

Journal of Petroleum

Science and Technology

1995, 17-21.

[17] Mottie H., Geology of Iran; Petroleum

Geology of Zagros, Geological Society of

Iran Publication, 1995, 37 (2), 1009.

[18] Nelles O., Nonlinear System Identification,

Springer-Verlag, New York, 2002, 13(4), 785.

[19] Rabiller P. and Ye S. J., The Iterative use

of Clustering and Modeling to Improve

Permeability Prediction, SPWLA 42th

Annual Logging Symposium, Houston,

Texas, 2001.

[20] Akbar M., Vissapragada B., Dhabi A.,

Alghamdi A. H., et al., A Snapshot of

Carbonate Reservoir Evaluation, Schlumbergeroil Field Review, 2000, 12(4), 20-41.

[21] Timur A., An Investigation of Permeability,

Porosity, and Residual Water Saturation

Relationships for Sandstone Reservoirs,

The Log Analysts, 1968, 9(4).

F. Khoshbakht

from Electric-Log Resistivity gradients, Oil

and Gas Journal, 1949, 17, 68-73.

[23] Wyllie M. R. J. and Rose W. D., Some

Theoretical Considerations Related to the

Quantitative Evaluation of the Physical

Characteristics of Reservoir Rock from

Electrical Log Data, Petroleum Transactions,

AIME, 1950, 2(4), 105-118.

[24] Ye S. J. and Rabiller P., A New Tool for

Electro-Facies Analysis: Multi-Resolution

Graph-Based Clustering, SPWLA 41st

Annual Logging Symposium, 2000.

[25] Zhang Y. J. and Lollback P. A., A

Methodology for Estimating Permeability

from Well Logs in a Formation of Complex

Lithology, paper SPE 37025, presented at

the SPE Asia Pacific Oil & Gas Conference,

Adelaide, Australia, 1996, 28-31.

2014 Research Institute of Petroleum Industry (RIPI)

90 |

http://jpst.ripi.ir

- Facimage Mrgc PaperUploaded byElok Galih Karuniawati
- Regression ChecklistUploaded byYixin Xia
- 0120-IdentifyingDuplicateRecordsinIDQ9Uploaded byRaghu Ka
- Fin SolUploaded bybaloch45
- biochemj00812-0107Uploaded byhaddig8
- 5832_chap1Uploaded byNikhil Sawant
- About the MS Regress PackageUploaded bygliptak
- Dividend Changes and Future ProfitabilityUploaded byBahulu Berinti 'Arashi
- Financial Leverage and Performance of Non-Financial Companies in Nairobi Securities Exchange in KenyaUploaded byIOSRjournal
- Statistics study notesUploaded bybakajin00
- TR-2003011Uploaded byjayson_rome
- evaluacion affordancesUploaded byVirginia Suárez
- tmpCC92.tmpUploaded byFrontiers
- Xl MinerUploaded byLoneli Costaner
- UploadUploaded byDan Lu
- Chapter17_3Uploaded byLan Anh Bùi
- Cluster Methods in SASUploaded byramanujsarkar
- paper_titUploaded byGaurav Gupta
- bettcher term paper apsy 607Uploaded byapi-162509150
- ewanUploaded byChristian Aquino
- researchUploaded byfortunatesanjib
- Segmentation FrankWyman SlidesUploaded byDanudear Daniel
- Camera Ready v1Uploaded by582d9c24
- metaUploaded byWinanda Tri Saputra
- 1836425.pdfUploaded byWulan Jessica
- Different Approaches to Defining CellsUploaded byMichael Thompson
- Teaching EconometricsUploaded bylucky_man612
- Advances in Data Stream MiningUploaded byduh1988
- Asseigment-6Uploaded byGoutham Kumar Allam
- 1-s2.0-S0306437915001349-mainUploaded byLê Kim Hùng

- OCTA ( STAR ) PVD A Different Approach of Image SteganopgraphyUploaded bySachin Thanekar
- ND W-CDMA Design Paper - Site Configuration - Iub - Ed16Uploaded byMohammed Mokhtar
- modified lesson planUploaded byapi-377051886
- MSDS RoundupUploaded byasrizal
- Manual De Parts Manual ST - LT - 270.pdfUploaded byRicardo Ramirez Estrello
- GateGenie csUploaded byKumar Gaurav
- Case Analysis of Huella Online TravelUploaded byPriyatham Kireeti
- ASTM Lab ManualUploaded byGovind
- NetApp Windows Storage ConsolidationUploaded byGangi Reddy
- zoomerang_wp_customersatUploaded bytmuqgjwh
- CSC162 Manual [1]Uploaded byGunji Venkata Srinivasa Babu
- Frankis Solutions Limited Thames Barrier Park East Silvertown Remedial MethodUploaded byJoanna Rodriguez
- Oral History TechniquesUploaded bysniad
- Good-times-bad-times.pdfUploaded byJohann Pourcelot
- EconomicsUploaded bySaranya Pugazhenthi
- Gyro to NMEA Convertor ManualUploaded byConstantinos Peppas
- Jack-Panels y JacksUploaded byjorgeestevez
- ch06Uploaded byAlrifai Ziad Ahmed
- EtherWAN EX27064-V0VC Data SheetUploaded byJMAC Supply
- Torque Convertor Ppt CtiUploaded byThiyagarajan Rajkumar
- 03 Modular Skid SystemsUploaded byjosethompson
- homoscedasticityUploaded bycallanant
- 2013 Timeline and Action ItemsUploaded bybygsky
- Siemens Tcpip EthernetUploaded byhiddenstar00
- City Beat News Announces Spectrum Award Winners for 2018Uploaded byPR.com
- Unibet Poker LogUploaded byAlex Iurascu
- ACL Ax 54x Tech BriefUploaded byrajsimhan
- Waterproofing in BasementsUploaded byMohsinShaikh
- S-300VUploaded bySohail Safeer
- Basic ThermodynamicsUploaded byJess Ryan Olayon